Tiptap extension command React DX

Purpose

Compile the Tiptap evidence that matters to Slate v2's extension ergonomics and React integration.

Strongest Evidence

• Tiptap's editor instance centralizes editor creation, extensions, content, editability, input rules, paste rules, editor props, and ProseMirror view setup.
• extensions package nodes, marks, attributes, global attributes, commands, events, and keyboard shortcuts.
• custom extensions add commands that appear on editor.commands and editor.chain().
• CommandManager builds single-command and chained-command APIs around one transaction.
• React performance docs recommend selective editor-state subscriptions and disabling transaction-wide rerenders.
• the composable API exposes editor context, loading states, bubble/floating UI, and selector hooks.

What To Steal

1. Extension ergonomics

Slate v2 should make feature packaging feel like:

• define extension
• add extension
• get methods/normalizers/handlers/UI outputs

Plate should not have to manually thread unrelated snippets across React providers, editor construction, and command handlers.

2. Command discoverability

Tiptap's command catalog is good DX. Slate v2 should expose discoverable editor methods and extension methods, but should keep the write lifecycle as editor.update.

3. Optional chain API

Tiptap's chain API is useful for composing multiple operations into one transaction. Slate v2 can offer optional sugar:

editor.chain().setNodes(props).wrapNodes(wrapper).run()

But chain().run() should be sugar over editor.update, not a second engine.

4. React selector posture

Tiptap docs correctly identify React rerender pressure around editor transactions. Slate v2 should go further:

• selector hooks should consume EditorCommit dirtiness
• urgent text paths should use live reads / DOM capability results
• unrelated React state should not rerender the editor body

5. Composable UI

Tiptap's composable React API is good product DX. Slate v2 / Plate should provide:

• provider
• content
• loading/ready state
• bubble/floating UI channels
• selector hooks
• toolbar helpers

What Not To Steal

• Do not require chain().focus().command().run() for ordinary toolbar commands.
• Do not make ProseMirror leakage the advanced-user default.
• Do not make React integration rely on isolating the editor from React.
• Do not use commands as a substitute for a safe read/update lifecycle.

Take For Slate v2

Tiptap is the product-DX benchmark, not the engine benchmark.

Slate v2 should use:

editor.update as lifecycle
primitive methods as power API
extension methods as product API
optional chain as sugar
selector hooks for React UI

This keeps Slate flexibility while giving Plate/Tiptap-level ergonomics.